mlsanityData to 7 Oct 2026

KEEP: A KV-Cache-Centric Memory Management System for Efficient Embodied Planning

arXiv 2602.23592 ↗First posted 27 Feb 2026cs.RO cs.AI cs.SEcited by 3

On the map: Efficient LLM inference and serving → KV cache optimization

Abstract (arXiv)

Memory-augmented Large Language Models (LLMs) have demonstrated remarkable capability for complex and long-horizon embodied planning. By keeping track of past experiences and environmental states, memory enables LLMs to maintain a global view, thereby avoiding repetitive exploration. However, existing approaches often store the memory as raw text, leading to excessively long prompts and high prefill latency. While it is possible to store and reuse the KV caches, the efficiency benefits are greatly undermined due to frequent KV cache updates. In this paper, we propose KEEP, a KV-cache-centric memory management system for efficient embodied planning. KEEP features 3 key innovations: (1) a Static-Dynamic Memory Construction algorithm that reduces KV cache recomputation by mixed-granularity memory group; (2) a Multi-hop Memory Re-computation algorithm that dynamically identifies important cross-attention among different memory groups and reconstructs memory interactions iteratively; (3) a Layer-balanced Memory Loading that eliminates unbalanced KV cache loading and cross-attention computation across different layers. Extensive experimental results have demonstrated that KEEP achieves 2.68x speedup with negligible accuracy loss compared with text-based memory methods on ALFRED dataset. Compared with the KV re-computation method CacheBlend (EuroSys'25), KEEP shows 4.13% success rate improvement and 1.90x time-to-first-token (TTFT) reduction. Our code is available on https://github.com/PKU-SEC-Lab/KEEP_Embodied_Memory.

Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Placed in KV cache optimization, one of 12 directions in Efficient LLM inference and serving. Builds on 20 of its 33 references that are in the map. First posted on arXiv 27 Feb 2026.

Builds on

20 of 33 references.

Closest ancestors · 6

Same direction, newest first

  1. 2025CentroidKV: Efficient Long-Context LLM Inference via KV Cache Clustering13 Jun
  2. 2025FreeKV: Boosting KV Cache Retrieval for Efficient LLM Inference19 May
  3. 2025ChunkKV: Semantic-Preserving KV Cache Compression for Efficient Long-Context LLM Inference1 Feb
  4. 2024ShadowKV: KV Cache in Shadows for High-Throughput Long-Context LLM Inference28 Oct
  5. 2024CacheBlend: Fast Large Language Model Serving for RAG with Cached Knowledge Fusion26 May
  6. 2023Efficient Memory Management for Large Language Model Serving with PagedAttention12 Sep

Other references · 14

  1. 2023AWQ: Activation-aware Weight Quantization for LLM Compression and Accelerationcited by 1,234
  2. 2025A-MEM: Agentic Memory for LLM Agentscited by 1,000
  3. 2020MPNet: Masked and Permuted Pre-training for Language Understandingcited by 973
  4. 2019ALFRED: A Benchmark for Interpreting Grounded Instructions for Everyday Taskscited by 773
  5. 2023Retroformer: Retrospective Large Language Agents with Policy Gradient Optimizationcited by 110
  6. 2025MemGen: Weaving Generative Latent Memory for Self-Evolving Agentscited by 99
  7. 2025KVFlow: Efficient Prefix Caching for Accelerating LLM-Based Multi-Agent Workflowscited by 56
  8. 2024EPIC: Efficient Position-Independent Caching for Serving Large Language Modelscited by 45
  9. 2024KARMA: Augmenting Embodied AI Agents with Long-and-short Term Memory Systemscited by 38
  10. 2025Look Back to Reason Forward: Revisitable Memory for Long-Context LLM Agentscited by 37
  11. 2025Robix: A Unified Model for Robot Interaction, Reasoning and Planningcited by 23
  12. 2025LightMamba: Efficient Mamba Acceleration on FPGA with Quantization and Hardware Co-designcited by 11
2 more
  1. 2025EfficientNav: Towards On-Device Object-Goal Navigation with Navigation Map Caching and Retrievalcited by 5
  2. 2025Breaking the Boundaries of Long-Context LLM Inference: Adaptive KV Management on a Single Commodity GPUcited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026Kamera: Unified Position-Invariant Multimodal KV Cache for Training-Free Reusecited by 3
  2. 2026Compute Globally, Materialize Locally: The Memory Contract of Sparse Event-KVcited by 1
  3. 2026BIO-MEMART: Biometric-Aware KV Cache Memory for Multi-User LLM Agentscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026AgenticCache: Cache-Driven Asynchronous Planning for Embodied AI Agents
  2. 2025FindingDory: A Benchmark to Evaluate Memory in Embodied Agents
  3. 2026RPMS: Enhancing LLM-Based Embodied Planning through Rule-Augmented Memory Synergy
  4. 2026Explore with Long-term Memory: A Benchmark and Multimodal LLM-based Reinforcement Learning Framework for Embodied Exploration
  5. 2025Exploratory Retrieval-Augmented Planning For Continual Embodied Instruction Following
  6. 20253DLLM-Mem: Long-Term Spatial-Temporal Memory for Embodied 3D Large Language Model
  7. 2026EmbodiedMemory-Bench: Benchmarking Embodied Memory for Long-Horizon Embodied Tasks
  8. 2026BrainMem: Brain-Inspired Evolving Memory for Embodied Agent Task Planning